
import plotly.graph_objects as go

categories = ['Whatever {}'.format(i) for i in range(1, 5)] + ['Data visualization']+["Whatever 1"]

scores = [2, 3, 4, 2, 4]
fig = go.Figure()

_ = fig.add_trace(go.Scatterpolar(
      r=scores+[2],
      theta=categories,
      name='Before this seminar',
      hoverinfo='r+name',
))
_ = fig.add_trace(go.Scatterpolar(
    r=scores[:-1]+[5]+[2],
    theta=categories,
    name='After this seminar',
    hoverinfo='r+name',
))

_ = fig.update_layout(
    width=400, height=400, font_size=14,
        polar=dict(radialaxis=dict(visible=True, range=[0, 5.5]),),
        margin=dict(l=100,r=100,b=0,t=0),
        showlegend=False,
        hoverlabel = dict(font=dict(size=16), namelength = -1),
)

fig.show()


# Relevant imports
import numpy as np
import pandas as pd

import plotly.graph_objects as go
from plotly.subplots import make_subplots

# Some sklearn imports for statistical methods
from sklearn.datasets import load_wine
from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import train_test_split
from sklearn.decomposition import PCA
from sklearn.linear_model import Lasso

# Some jupyter-notebook related imports/settings
from plotly.offline import init_notebook_mode, plot
from IPython import display
from IPython.display import IFrame
from IPython.core.interactiveshell import InteractiveShell
init_notebook_mode(connected=False)
InteractiveShell.ast_node_interactivity = "all"


empty_fig = go.Figure()
empty_fig.show()


# Create random sample data
scatter_data = np.random.rand(10)

# Create figure
scatter_fig = go.Figure()

# Add scatter-plot
# Assignment to _ variable similar to redirection to /dev/null in bash. Avoids output of return value.
_ = scatter_fig.add_trace(
        go.Scatter(
            y=scatter_data,
    #       x=list(range(len(scatter_data))),
            mode='markers',
        )
    )

# Show figure
_ = scatter_fig.show()


# Let's create our data
x = np.arange(0, 4*np.pi+np.pi/100, np.pi/100)
y = np.sin(x)

# Create figure
sine_fig = go.Figure()

# Add scatter-plot
_ = sine_fig.add_trace(
    go.Scatter(
        x=x,
        y=y,
        mode='lines',
    )
)

# Show figure
_ = sine_fig.show()


# Define figure layout
sine_layout = go.Layout(
    
    width=800, height=400,
    title=dict(text='Simple sine function', x=0.55,
               font_size=20,                                 # Magic underscore notation
              ),
#   title_text='Simple sine function',

    margin=dict(l=80, r=0, b=50, t=100,),

    xaxis=dict(
        title='x', title_font=dict(size=18,),
        tickfont=dict(size=18,), 
        tickmode='linear', tick0=0, dtick=np.pi,
        gridcolor='black',
    ),
)

# Apply layout to figure
_ = sine_fig.update_layout(sine_layout)

# Apply layout-update to certain axes only
_ = sine_fig.update_yaxes(title=dict(text='sin(x)', font_size=18, font_color='blue'),
        range=[-1.05, 1.05],
        tickmode='linear', tick0=0, dtick=0.5, tickfont=dict(size=18,),
        gridcolor='black', zerolinecolor='black'
    )

# Show figure
_ = sine_fig.show()


# Load data
wine_data = load_wine()
wine_df = pd.DataFrame(wine_data.data, columns=wine_data.feature_names)

wine_df.head()
print("Shape of wine-dataframe: {}".format(wine_df.shape))

Shape of wine-dataframe: (178, 13)


# Create Figure
wine_fig = go.Figure()

# Add wine-data as Bar-chart
_ = wine_fig.add_trace(
    go.Bar(
        y=wine_df.sort_values(by='alcohol', ascending=False).iloc[:20]['alcohol'],
    )
)

# Define layout
wine_layout = go.Layout(
    width=1000, height=350,
    font_size=10,
    title=dict(text='Alcohol content of 20 strongest wines in dataset', x=0.55,
               font_size=20),
    
    margin=dict(l=80, r=0, b=0, t=100,),
    
    xaxis=dict(
        title='rank', title_font=dict(size=18,),
        tickmode='linear', tick0=2, dtick=1,
        tickfont=dict(size=18,),
    ),
    
    yaxis=dict(
        title=dict(text='% Alcohol', font_size=18,),
        range=[14, 15],
        tickfont=dict(size=18,),
    )
)

# Update layout
_ = wine_fig.update_layout(wine_layout)

# Show figure
_ = wine_fig.show()


# Let's create some random data
heat_data = np.random.randn(10, 10)

# Create figure
heat_fig = go.Figure()

# Add heatmap-trace
_ = heat_fig.add_trace(go.Heatmap(z=heat_data,),)

# Define layout
heat_layout = go.Layout(
    width=500, height=500,
    title=dict(text='Basic heatmap', x=0.43, font_size=20,),
    margin=dict(l=0, r=0, b=0, t=100,),
)

# Update layout
_ = heat_fig.update_layout(heat_layout)

# Show figure
_ = heat_fig.show()


# Let's create some random sample data
metals = ["Ni", "Pd", "Pt"]
energy_data = np.random.uniform(low=0, high=3, size=(10,3))

# Let's set our "ground states"
energy_data[0,:] = 0

# Let's introduce some outliers
energy_data[-1, 0] = energy_data[-1, 0]+6
energy_data[-1, 1] = energy_data[-1, 1]+4
energy_data[-1, 2] = energy_data[-1, 2]-4

# DataFrame creation
energy_df = pd.DataFrame(data=energy_data, columns=metals)

energy_df.head(10)


# Create custom hovertexts
hover_texts = ["ground state"]+["all good"]*8+["messed up POTCAR"] + \
              ["ground state"]+["all good"]*8+["not converged"] + \
              ["ground state"]+["all good"]*8+["wrong k-point sampling"]

hover_texts = [hover_text+'<br>Directory {}'.format(i) for i, hover_text in enumerate(hover_texts)]
hover_texts = np.reshape(hover_texts, (3, 10))
hover_texts.shape
print(hover_texts[0])

(3, 10)

['ground state<br>Directory 0' 'all good<br>Directory 1'
 'all good<br>Directory 2' 'all good<br>Directory 3'
 'all good<br>Directory 4' 'all good<br>Directory 5'
 'all good<br>Directory 6' 'all good<br>Directory 7'
 'all good<br>Directory 8' 'messed up POTCAR<br>Directory 9']


# Create Figure
energy_fig = go.Figure()

metal_colors = ["red", "green", "blue"]

# Add one trace per metal for more control
for imetal, metal in enumerate(metals):
    _ = energy_fig.add_trace(
            go.Scatter(
                x=[imetal]*len(energy_df[metal]),
                y=energy_df[metal],
                mode='markers', marker_size=8, name=metal,
                
                marker_color=metal_colors[imetal],
                hoverinfo='x+y+text',
                text=hover_texts[imetal],
            ),
    )
#   print(hover_texts[imetal])

# Zero-reference line
_ = energy_fig.add_trace(
        go.Scatter(
            x = [0, 2],
            y = [0, 0],
            mode='lines', line_color='grey', line_width=2,
            showlegend=False, hoverinfo='skip',
        ),
)

# Update figure layout
energy_layout = go.Layout(
    
    width=700, height=500,
    font_size=10,
    title=dict(text='Some calculated energies', x=0.5, font_size=20),
    
    margin=dict(l=80, r=0, b=0, t=100,),
    
    xaxis=dict(
        title='Metal', title_font=dict(size=18,),
        tickmode = 'array', tick0 = 1, dtick = 1,
        tickfont=dict(size=18,),
        tickvals=[0, 1, 2],
        ticktext=metals,
    ),
    
    yaxis=dict(
        title=dict(text='DFT energy / eV', font_size=18),
        tickfont=dict(size=18,),
    )
)

_ = energy_fig.update_layout(energy_layout)

# Plot and save figure
_ = energy_fig.show()
_ = plot(energy_fig, filename='./dft_energies.html', auto_open=False)


# Let's create some sample MD data
md_mos = np.random.uniform(low=0, high=0.3, size=(500,))
md_mos[100:200] += 1
md_mos[200:220] += 1.5
md_mos[220:300] += 2
md_mos[300:400] += 1
md_mos[400:450] += 0.5

# Scale M-OS into values between 0 and 1 to be used in rgba-color specification
mos_min, mos_max, mos_mean = np.min(md_mos), np.max(md_mos), np.mean(md_mos)
mos_colors = [np.log(mos+1) for mos in md_mos]
mos_colors = [(mos-mos_min)/(mos_max-mos_min) for mos in mos_colors]

# Create figure
md_fig = go.Figure()

# Add one line-trace for each data-point
for imos, mos in enumerate(md_mos):
    
    # Add vertical line for each MD-step
    _ = md_fig.add_trace(go.Scatter(
            x=[imos, imos],
            y=[mos_min, mos_max],
            mode='lines', text=str(mos),
            line=dict(color='rgba({}, {}, {}, {})'.format(0, 0, 255*(1-mos_colors[imos]), 1), width=3),
            hoverinfo='x+text', showlegend=False,
            )
    )

# Add overall Pt-OS line
_ = md_fig.add_trace(go.Scatter(
        y=md_mos,
        mode='lines', text=md_mos,
        line=dict(color='white', width=2, shape='spline'),
        hoverinfo='skip', showlegend=False,
        )
    )

# Define custom colorscale legend via heatmap
color_array = np.linspace(min(mos_colors), max(mos_colors), 51)
color_list = ['rgba({}, {}, {}, {})'.format(0, 0, 255*(1-color_val), 1) for color_val in color_array]
_ = md_fig.add_trace(go.Heatmap(
        z=[np.linspace(mos_min, mos_max, 51)],
        colorscale=[color for color in color_list],
        )
    )

# Add text annotation
_ = md_fig.add_annotation(
    xanchor="left", yanchor="top",
    x=0.02*len(md_mos), y=0.99*max(md_mos),
    text="Pt @ 600K",
    font=dict(color='white', size=24),
    showarrow=False,
)

# Define layout
md_layout = go.Layout(
    title=dict(text="Pt-OS during MD simulation", font_size=18, x=0.5,),
    width=900, height=450, margin=dict(l=80, r=0, b=0, t=50,),
    autosize=True,
    yaxis=dict(title='Pt-OS', side='left', range=[mos_min, mos_max]),
    xaxis=dict(title='MD step', range=[0, len(md_mos)],),
    hovermode='x unified', hoverlabel={'namelength': -1},
)

# Update layout
_ = md_fig.update_layout(md_layout)

# Show and save figure
_ = md_fig.show()
_ = plot(md_fig, filename='./md_os.html', auto_open=False)

del md_fig


# Let's define some random data
feature_data = np.random.uniform(low=0, high=3, size=(100,8))
feature_names = ['feature {}'.format(i) for i in range(8)]
df = pd.DataFrame(data=feature_data, columns=feature_names)
df.head()


# Data standardization
scaler = StandardScaler().fit(feature_data)
x_scaled = scaler.transform(feature_data)

# Run PCA
n_components=4
pca = PCA(n_components=n_components)
x_transformed = pca.fit_transform(x_scaled)

pca_headers = ['PC {}'.format(pci) for pci in range(x_transformed.shape[1])]

# Define df with feature-weights of each PC
weight_df = pd.DataFrame(
    data=pca.components_.transpose(),
    columns=pca_headers,
).abs()


# Define figure with 2 vertically stacked subplots
pca_fig = make_subplots(rows=2, subplot_titles=["Cumulative variance", "Weights > 0.3"],
                        vertical_spacing=0.1, row_heights=[0.1, 0.9], shared_xaxes=True,)

# Add bar chart of cumulative variance to first subplot
cumulative_variance = list(np.cumsum(pca.explained_variance_ratio_))
_ = pca_fig.add_trace(go.Bar(
        x=['PC '+ str(i) for i in range(n_components)],
        y=cumulative_variance,
        showlegend=False, textposition='auto',
        text=[round(y, 3) for y in cumulative_variance],
        hoverinfo='x+y',
    ), 
    row=1, col=1,
)

# Add weight heatmap to second subplot
_ = pca_fig.add_trace(go.Heatmap(
        x=['PC '+ str(i) for i in range(n_components)],
        y=feature_names,
        z=weight_df[weight_df > 0.3].values,
        xgap=1, ygap=1, colorbar=dict(thickness=20, ticklen=3, len=0.8, y=0.4),
        hoverinfo='x+y+z',
    ),
    row=2, col=1,
)

# Define global layout
pca_layout = go.Layout(
    title=dict(text='Principal component analysis (PCA)', x=0.5),
    width=800, height=600, font=dict(size=12),
    margin=dict(l=0, r=0, b=0, t=100,),
)

# Update global layout
_ = pca_fig.update_layout(pca_layout)

# Show figure
_ = pca_fig.show()
_ = plot(pca_fig, filename='./pca.html', auto_open=False)


# Define data
lasso_target = df["feature 0"]  - \
             2*df["feature 1"]  + \
     3*np.sqrt(df["feature 2"]) - \
     .1*np.exp(df["feature 3"])

x = feature_data
y = lasso_target

# Data splitting and standardization
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2, random_state=0, shuffle=False)
train_scaler = StandardScaler().fit(x_train)
x_train_scaled = train_scaler.transform(x_train)
x_test_scaled  = train_scaler.transform(x_test)

# Vary alpha-parameter of Lasso-model
alphas = np.linspace(start=float('1e-5'), stop=float('2'), num=1000)
coefs = []
for alpha in alphas:
    model = Lasso(alpha)
    _ = model.fit(x_train_scaled, y_train)
    coefs.append(model.coef_)


# Create figure
lasso_feat_fig = go.Figure()

# Plot course of Features
for ifeat, feat in enumerate(feature_names):
    _ = lasso_feat_fig.add_trace(go.Scatter(
        x=alphas,
        y=[coef[ifeat] for coef in coefs],
        mode='lines', line=dict(width=3), showlegend=True, name=feat,
    ),
)

lasso_feat_layout = go.Layout(
    width=1000, height=400, font=dict(size=12), margin=dict(l=0, r=0, b=0, t=50,),
    title=dict(text='LASSO - Features', x=0.47,),
    xaxis=dict(title=r'$\alpha$', range=[0, max(alphas)],),
    yaxis=dict(title='Weights',),
    
    hoverlabel = {'namelength': -1}, hovermode='x unified',
)

_ = lasso_feat_fig.update_layout(lasso_feat_layout)

# Save and plot Figure
_ = plot(lasso_feat_fig, filename='./lasso.html', auto_open=False)
_ = lasso_feat_fig.show()


print(type(energy_fig))
print('='*42)
print(energy_fig)

<class 'plotly.graph_objs._figure.Figure'>
==========================================
Figure({
    'data': [{'hoverinfo': 'x+y+text',
              'marker': {'color': 'red', 'size': 8},
              'mode': 'markers',
              'name': 'Ni',
              'text': array(['ground state<br>Directory 0', 'all good<br>Directory 1',
                             'all good<br>Directory 2', 'all good<br>Directory 3',
                             'all good<br>Directory 4', 'all good<br>Directory 5',
                             'all good<br>Directory 6', 'all good<br>Directory 7',
                             'all good<br>Directory 8', 'messed up POTCAR<br>Directory 9'],
                            dtype='<U38'),
              'type': 'scatter',
              'x': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
              'y': array([0.        , 0.72728177, 0.03645188, 2.4359331 , 0.64700342, 1.19968771,
                          2.49073165, 0.51384806, 2.89752347, 8.19224525])},
             {'hoverinfo': 'x+y+text',
              'marker': {'color': 'green', 'size': 8},
              'mode': 'markers',
              'name': 'Pd',
              'text': array(['ground state<br>Directory 10', 'all good<br>Directory 11',
                             'all good<br>Directory 12', 'all good<br>Directory 13',
                             'all good<br>Directory 14', 'all good<br>Directory 15',
                             'all good<br>Directory 16', 'all good<br>Directory 17',
                             'all good<br>Directory 18', 'not converged<br>Directory 19'],
                            dtype='<U38'),
              'type': 'scatter',
              'x': [1, 1, 1, 1, 1, 1, 1, 1, 1, 1],
              'y': array([0.        , 0.26619657, 0.74241565, 0.38315368, 1.82520905, 2.22683916,
                          2.37178794, 0.65535811, 1.1067929 , 6.57749848])},
             {'hoverinfo': 'x+y+text',
              'marker': {'color': 'blue', 'size': 8},
              'mode': 'markers',
              'name': 'Pt',
              'text': array(['ground state<br>Directory 20', 'all good<br>Directory 21',
                             'all good<br>Directory 22', 'all good<br>Directory 23',
                             'all good<br>Directory 24', 'all good<br>Directory 25',
                             'all good<br>Directory 26', 'all good<br>Directory 27',
                             'all good<br>Directory 28', 'wrong k-point sampling<br>Directory 29'],
                            dtype='<U38'),
              'type': 'scatter',
              'x': [2, 2, 2, 2, 2, 2, 2, 2, 2, 2],
              'y': array([ 0.        ,  1.81461338,  1.65425372,  1.09956177,  2.47774639,
                           1.76447392,  2.46019888,  0.53385378,  0.92504454, -3.0185493 ])},
             {'hoverinfo': 'skip',
              'line': {'color': 'grey', 'width': 2},
              'mode': 'lines',
              'showlegend': False,
              'type': 'scatter',
              'x': [0, 2],
              'y': [0, 0]}],
    'layout': {'font': {'size': 10},
               'height': 500,
               'margin': {'b': 0, 'l': 80, 'r': 0, 't': 100},
               'template': '...',
               'title': {'font': {'size': 20}, 'text': 'Some calculated energies', 'x': 0.5},
               'width': 700,
               'xaxis': {'dtick': 1,
                         'tick0': 1,
                         'tickfont': {'size': 18},
                         'tickmode': 'array',
                         'ticktext': [Ni, Pd, Pt],
                         'tickvals': [0, 1, 2],
                         'title': {'font': {'size': 18}, 'text': 'Metal'}},
               'yaxis': {'tickfont': {'size': 18}, 'title': {'font': {'size': 18}, 'text': 'DFT energy / eV'}}}
})


energy_fig.show("json")


# # Static export requires some dependencies, e.g. psutil, plotly-orca
_ = energy_fig.write_image('./dft_energies.png')
_ = energy_fig.write_image('./dft_energies.svg')
_ = energy_fig.write_json('./dft_energies.json')

# Export as interactive figures in html-format
_ = plot(energy_fig, filename='./dft_energies.html', auto_open=False)
_ = energy_fig.write_html('./dft_energies.html')

# Import of interactive html-figure into jupyter-notebook
# del energy_fig
IFrame("./dft_energies.html", width=750, height=550)


error_layout = go.Layout(
    width=400, height=400,
#     abc='abc',
    xaxis=dict(
#           abc='abc',
#          ticks='abc',
    )
)


pd.options.plotting.backend = "matplotlib"

df = pd.DataFrame(dict(a=[1,3,2], b=[3,2,1]))
df.head()
fig = df.plot()


pd.options.plotting.backend = "plotly"

fig1 = df.plot()
fig1.show()


fig2 = df.plot.bar()
fig2.show()

	alcohol	malic_acid	ash	alcalinity_of_ash	magnesium	total_phenols	flavanoids	nonflavanoid_phenols	proanthocyanins	color_intensity	hue	od280/od315_of_diluted_wines	proline
0	14.23	1.71	2.43	15.6	127.0	2.80	3.06	0.28	2.29	5.64	1.04	3.92	1065.0
1	13.20	1.78	2.14	11.2	100.0	2.65	2.76	0.26	1.28	4.38	1.05	3.40	1050.0
2	13.16	2.36	2.67	18.6	101.0	2.80	3.24	0.30	2.81	5.68	1.03	3.17	1185.0
3	14.37	1.95	2.50	16.8	113.0	3.85	3.49	0.24	2.18	7.80	0.86	3.45	1480.0
4	13.24	2.59	2.87	21.0	118.0	2.80	2.69	0.39	1.82	4.32	1.04	2.93	735.0

	Ni	Pd	Pt
0	0.000000	0.000000	0.000000
1	0.727282	0.266197	1.814613
2	0.036452	0.742416	1.654254
3	2.435933	0.383154	1.099562
4	0.647003	1.825209	2.477746
5	1.199688	2.226839	1.764474
6	2.490732	2.371788	2.460199
7	0.513848	0.655358	0.533854
8	2.897523	1.106793	0.925045
9	8.192245	6.577498	-3.018549

	feature 0	feature 1	feature 2	feature 3	feature 4	feature 5	feature 6	feature 7
0	0.115450	1.730281	2.158649	1.895502	0.660554	0.617611	1.467029	2.962683
1	2.178489	1.172428	0.466978	2.052390	1.512839	2.445880	0.552615	1.873959
2	2.190881	2.526951	2.872476	1.626631	0.697366	1.672346	2.230905	2.405939
3	1.407376	2.952416	2.192189	1.804616	2.391801	1.384519	1.832575	2.151778
4	2.297769	0.335774	0.305688	1.954677	1.503372	0.800529	1.676376	1.152252

1. Introduction

2. Creating and customizing plots

2.1 Empty canvas: `go.Figure()`

2.2 Simple scatter plot with `add_trace()` and `go.Scatter()`

2.3 Simple line plot with `add_trace()` and `go.Scatter()`

2.4 Figure customization with `go.Layout()` and `update_layout()`

2.5 Simple bar chart with `go.Bar()`

2.6 Simple heatmap with `go.Heatmap()`

3. Some exemplary use cases

3.1 Plotting of DFT energies

3.2 Visualization of Molecular Dynamics results

3.3 Applications in statistical learning - PCA and LASSO

4. Miscellaneous tips and tricks

4.1 Internal workings of the Figure class

4.2 Saving, exporting and importing of figures

4.3 Instant documentation in Error Tracebacks

4.4 Direct plotting from Pandas-DataFrames

4.5 Further resources

1. Introduction

2. Creating and customizing plots

2.1 Empty canvas: go.Figure()

2.2 Simple scatter plot with add_trace() and go.Scatter()

2.3 Simple line plot with add_trace() and go.Scatter()

2.4 Figure customization with go.Layout() and update_layout()

2.5 Simple bar chart with go.Bar()

2.6 Simple heatmap with go.Heatmap()

3. Some exemplary use cases

3.1 Plotting of DFT energies

3.2 Visualization of Molecular Dynamics results

3.3 Applications in statistical learning - PCA and LASSO

4. Miscellaneous tips and tricks

4.1 Internal workings of the Figure class

4.2 Saving, exporting and importing of figures

4.3 Instant documentation in Error Tracebacks

4.4 Direct plotting from Pandas-DataFrames

4.5 Further resources

2.1 Empty canvas: `go.Figure()`

2.2 Simple scatter plot with `add_trace()` and `go.Scatter()`

2.3 Simple line plot with `add_trace()` and `go.Scatter()`

2.4 Figure customization with `go.Layout()` and `update_layout()`

2.5 Simple bar chart with `go.Bar()`

2.6 Simple heatmap with `go.Heatmap()`